I Will Be Censored for Explaining It to You! | Robert Kiyosaki

Overview

This is going to happen this year. So, get ready, there are things looming with the implementation of Artificial Intelligence that are going to give you goosebumps next year because there is a lot of transformation underway in this domain. In fact, that has been the case particularly in the last few months. It’s almost unimaginable; you’re going to see things that you simply won’t be able to believe.

Introduction to Chat GPT

How many of you would applaud if I asked what chat GPT is? Well, not many. I’m going to tell you what chat GPT is so you know, because you need to know. And I don’t know what kind of technological revolution this is; it’s on the level of the Gutenberg press, it’s something like that. This is something big. This artificial intelligence system is a general model of language processing. It was released a week ago, a week and a half ago, and I went and interacted with it. It’s an artificial intelligence system basically trained on a massive corpus of spoken words and text. Essentially, it derives its models from the world of human speech analysis. It’s not yet using real-world data, but that will undoubtedly happen in the next year. Chat GPT analyzes a huge body of text that keeps growing, and it’s already quite sophisticated.

Chat GPT’s Versatility

Daoist morality, ethics, and the ethics described in the Sermon on the Mount, and he just nailed it brilliantly again. It took about 3 seconds. There was a computer engineer who intended to work for Tesla. He asked Chat GPT, he said, “Look, I work for Elon Musk, but I haven’t been doing much over the past week. I need you to write me 10 scripts about what I probably would have done as an engineer on Twitter. What 10 things did I do last week that were productive and valuable? Oh, and if you don’t mind, write me the computer code that accompanies each project.” And it also did that in 3 seconds. And the computer code works fine. So, that’s already there. On the other hand, a university professor did this, thought, “Oh, this is interesting. Any student will be able to write any essay on any topic with Chat GPT.” And by the way, someone gave it an SAT exam and got almost as good a score as the average student at a public university. That works well. So, it’s that smart. It’s basically an IQ test. He said, “Write me an essay.” He gave it a topic, wrote the essay, said, “Now grade us. Automate the students; we should be able to automate the teachers too.” It provided a thorough, comprehensive analysis of its own essay with a grade. Someone else asked it to write the script and describe the characters of the next $900 million Hollywood blockbuster, and Bank characterized the plot at that moment. Someone else gave it the descriptions of the actors and told it to generate realistic virtual images for each actor, and all the AI systems could do that. Now, I’m going to tell you what’s going to happen next. This is going to happen this year. So, get ready. Now we have an AI model that can extract a model of the real world through language composition, alright? And it’s smarter than you, and it’s going to be way smarter than you in 2 years. So, get ready for that too. But it’s not as smart yet because right now it’s just a humanities professor. It doesn’t test its linguistic knowledge in the real world. That’s what a scientist does. They come up with a linguistically predicated theory and then they throw it against the world and see if it holds. That’s when the world says whether your linguistic construction is valid or not. But the new AI systems will be able to extract patterns from the world itself, from images and such, so they’ll be able to test their linguistic constructions in the real world, just like scientists. And the more advanced models will use texts, images, and actions too because they’ll build a model of human action. This is all coming for next year, so hold on to your hats, ladies and gentlemen, because as my friend Jonathan said, “The Giants are going to walk the Earth once again, and we’re probably going to live to see it.” So, anyway, in terms of our rights, it’s not yet a bad development. Well, we won’t see it, that is, Elon Musk is one of the things he’s working on. He thinks the world will be controlled by whoever produces the most functional and fastest AI system because it’s going to be an advantage to be the first. And one of the things Musk has been working on for a long time is distributed AI systems so that you’ll have your own AI to protect yourself against, say, Google’s AI or the CSP AI because you can bet your hat they’re working on this as fast as they can. It seems necessary for us humans to see the world through history, in fact, when we describe the structure that governs our action and our perception. That’s the story, and so we’ve been trying to decipher, I would say to a certain extent on the religious front, what might be the deepest stories, and I’m very curious about.

Cognitive Processing and AI Development

The fact that we perceive the world through history is what humans do, and that seems to be a fundamental part of our cognitive architecture and cognitive architecture in general, according to some of the world’s top neuroscientists. I’m curious, and I know Jim is interested in cognitive processing and in building systems that in some way seem to work analogously to how our brains work. I’m curious about the notion that we have of seeing the world through history and what’s happening in the field of AI. It has taken us by surprise what has happened in the last 5 years; the speed at which models have started to do interesting and intelligent things. It has been estimated that human brains perform between 10 and 18 operations per second, which seems like a lot to me. It’s a trillion, a trillion operations per second. A small computer like your phone’s processor probably does 10 billion operations per second. If you use a GPU, maybe it reaches 100 billion or so. And the large modern AI computers like the ones used by OpenAI or Google do between 10 and 16 operations per second, maybe a little more. They’re just about to have the raw computational capacity of a human brain. And by the way, this could be completely wrong; our understanding of how the human brain computes could be wrong. But many people have made estimates based on the number of neurons, the number of connections, the firing speed of neurons, the number of operations that seems to imply the firing of a neuron. I mean, estimates vary by a couple of orders of magnitude. But when our computers got fast enough, we started to build things called language models and image models that do quite remarkable things. What have you seen in the last few years that is indicative of this change that you qualify as revolutionary? What are computers doing now that has surprised you due to this increase in speed? Yes, you can have a language model that reads a 200,000-word book and summarizes it fairly accurately. So it can extract the essence accurately. Could it be with fiction? Yes, yes, and I’ll introduce you to a friend who took a language model, changed it, fine-tuned it with Shakespeare, and used it to write scripts that are quite good. These kinds of things are really interesting. We were talking about this a little earlier when computers do calculations, you know, a program will say the sum of a + b equals c. The computer performs those operations in representations of information, ones and zeros. It doesn’t understand them at all. Ultimately, a computer doesn’t understand it. But what we call a language model translates information like words, images, and ideas into a space where for the program, the ideas and the operation it performs on them are essentially the same thing, right? A language model can produce words and then use those words as inputs, and it seems to have an understanding of what those words are, which is very different from how our computer operates with data. I’m very curious about language models. My feeling, at least in part of how we understand history, is that maybe we put that we are watching a movie; we have an idea of the character’s goals, then we see how that character perceives the world, and in a certain

Understanding Stories and Emotional Engagement

In a sense, we adopt their goals, we identify with the character, then we represent a panoply of emotions and motivations in our body because we now inhabit that space of goals, and we understand the character as a consequence of mimicking them with our own physiology. There are computers that can summarize the essence of a story but they don’t have that underlying physiology. Well, first of all, it’s a theory that your physiology has anything to do with it; you could understand the character’s goals and then engage in the details of the story. And then you would be predicting the path of the story and also having expectations and hopes for the story. Yes, and a good story takes you for a ride because it teases you with some of the things you expect but also with unexpected things. Yes, and that can create emotions. That’s right. So, in an AI model because you can easily have a set of goals, you have your personal goals. And then when you see the story, it has its goals. If you put them together, how many goals are the story’s goals and your goals? Hundreds, thousands, they’re small numbers, right? So you have the story, the AI model can also predict the story as well as you. That’s something that seems mysterious to me. As the story progresses, you can see the error between what was predicted and what actually happened, and then iterate on that. That’s what you would call emotional excitement, perception, anxiety. A big part of what anxiety seeks is the discrepancy. In fact, some of those states manifest in your body because they trigger hormonal cascades, right? But you can also scan your brain and see those things moving, right? And you know the AI model can have an error function and see the difference between what was expected and what wasn’t. You could call that an emotional state if you want to call it that. And this is speculative. No, I don’t think it’s accurate. We can make an AI model that could predict the outcome of a story probably better than the average person. Some people are really good, you know, they’re really well educated about stories or they know the genre or something. But you know what I’m seeing these days in terms of the model’s capabilities is that if you start describing it too much, it will make sense for a while but gradually it will stop making sense. Yet this is possible. It’s just the model’s capability at the moment, and the model is not well enough grounded in the foundations to define goals in reality or something like that when we look at how fast things are moving currently. And as it progresses, if you fast forward 10 years or see the relationship between the AI systems being built and humans, what would you imagine or could you imagine? Well, I can. As I said, I’m a computer guy, and I’m seeing this with, let’s say, a certain fascination. Also, I mean this last because according to Ray Hardwell, progress is accelerating, right? Yes, we have this idea that years of progress are 20 years but you know the last 20 years of progress were 20 years. The next 20 years will probably become from 5 to 10, right? True, true, true, and you can really feel it happening at a certain level causes social stress regardless of whether it’s AI or Amazon deliveries. There are many things that add to the stress of all this, but there is progress. It’s an extension of human capability, and then there is this progress that I’m hearing about the way you’re describing it that seems to be an inevitable progress towards the creation of something that

Human Desire and AI Development

It is more powerful than you, yes, and what is that? I don’t even understand that impulse. What is that impulse to create something that could supplant… Look, the average person in the world, yes, the average person already exists in this world because the average person is halfway up the human hierarchy. There are already many people more powerful than any of us could be, they could be smarter, they could be richer, they could be better connected. We live in a world where very few people are at the top of something, yes, so that’s already one thing. Basically, it’s like the impulse to make someone a superstar or the impulse to elevate someone above you. It’s the same thing that’s driving the creation of these ultra-powerful machines because we have that, we have the impulse to elevate. It’s like when we see a rock star we like, people want to submit to that, they want to dress like them, they want to elevate them above, like an example to follow, something to submit to. You see it with leaders, you see it in the political world, and in teams, you see it in sports teams, the same. Well, we’ve always tried to build things that are beyond us. It’s about, we’re building a God. That’s the impulse that’s pushing someone towards it because that’s what I hear with what you’re describing, Jin, I hear something that is extremely dangerous, right? It sounds extremely dangerous for the very existence of humans. However, I see humans acting and moving in that direction almost unable to stop it. I think it’s unstoppable, well, that’s one of the things we’ve also been talking about. I asked Jim directly about the hypothetical danger associated with this, why not stop doing it? Well, part of the answer is ambivalence about the outcome, but also that it’s not obvious at all that in a certain sense, it’s stopping. I mean, it’s the result of the cumulative action of a lot of people who are pushing this and even if you eliminate one player, even if it’s a key player, the probability that you’ll do more than slow it down is infinitely decimal, it’s quite low because there’s also a huge reward for those who succeed, it’s also set up that way, people know at least until the AI takes control or whatever, that whoever is in line towards increasing the power of AI will get great rewards. So it’s about cognitive acceleration, right? One of the things you might think might be dangerous in the field of AI is that if we optimize the way we interact with our electronic devices is to capture short-term tension, right? Because there’s a difference between getting what you want right now in the moment and getting what you need in a more mature sense over a reasonable period of time. And one of the things that seems to be happening on the internet and that I think is driven by the development of AI systems is that we’re being bombarded by systems that parasitize our short-term attention at the expense of long-term tension. If AI systems emerge to optimize tension grip, it’s not obvious to me that they will optimize tension that works in the medium and long term. Conceivably, they could maximize something like sporadic whim. Yes, because all virality is based on this, all social media, all of them are based on this reduction of tension, on this reduction of the desire to become your best version, they like to click on all these things. Yes, but that’s something that for reasons that are somewhat perplexing and sometimes drive me crazy, business models revolve around many of these interfaces. You know, the user is the product there, advertisers are trying to grab your attention. But that’s something that culture regulates. I see a lot of people using it and working on a variety of different sites. The cost is going to go down so fast that pretty soon you’ll have your own AI app that you’ll use to mediate the internet to remove the endless flow of ads, and you’ll be able to say, “Okay, this is the objective story, here are the 15 stories today, and this has been manipulated this way, this is being manipulated this other way,” and you can wonder which one resembles more to the real story. The curious thing is that the information is widely distributed and has many inputs, it’s very difficult to falsify it completely, so right now a story can go through a major media outlet and the narrative can be controlled, everyone gets the false story, but if the media is distributed among a billion people who interact in some useful way, there would be a real network, there would be.

Concerns about Superintelligent AI

It is more powerful than you, yes, and what is that? I don’t even understand that impulse. What is that impulse to create something that could supplant… Look, the average person in the world, yes, the average person already exists in this world because the average person is halfway up the human hierarchy. There are already many people more powerful than any of us could be, they could be smarter, they could be richer, they could be better connected. We live in a world where very few people are at the top of something, yes, so that’s already one thing. Basically, it’s like the impulse to make someone a superstar or the impulse to elevate someone above you. It’s the same thing that’s driving the creation of these ultra-powerful machines because we have that, we have the impulse to elevate. It’s like when we see a rock star we like, people want to submit to that, they want to dress like them, they want to elevate them above, like an example to follow, something to submit to. You see it with leaders, you see it in the political world, and in teams, you see it in sports teams, the same. Well, we’ve always tried to build things that are beyond us. It’s about, we’re building a God. That’s the impulse that’s pushing someone towards it because that’s what I hear with what you’re describing, Jin, I hear something that is extremely dangerous, right? It sounds extremely dangerous for the very existence of humans. However, I see humans acting and moving in that direction almost unable to stop it. I think it’s unstoppable, well, that’s one of the things we’ve also been talking about. I asked Jim directly about the hypothetical danger associated with this, why not stop doing it? Well, part of the answer is ambivalence about the outcome, but also that it’s not obvious at all that in a certain sense, it’s stopping. I mean, it’s the result of the cumulative action of a lot of people who are pushing this and even if you eliminate one player, even if it’s a key player, the probability that you’ll do more than slow it down is infinitely decimal, it’s quite low because there’s also a huge reward for those who succeed, it’s also set up that way, people know at least until the AI takes control or whatever, that whoever is in line towards increasing the power of AI will get great rewards. So it’s about cognitive acceleration, right? One of the things you might think might be dangerous in the field of AI is that if we optimize the way we interact with our electronic devices is to capture short-term tension, right? Because there’s a difference between getting what you want right now in the moment and getting what you need in a more mature sense over a reasonable period of time. And one of the things that seems to be happening on the internet and that I think is driven by the development of AI systems is that we’re being bombarded by systems that parasitize our short-term attention at the expense of long-term tension. If AI systems emerge to optimize tension grip, it’s not obvious to me that they will optimize tension that works in the medium and long term. Conceivably, they could maximize something like sporadic whim. Yes, because all virality is based on this, all social media, all of them are based on this reduction of tension, on this reduction of the desire to become your best version, they like to click on all these things. Yes, but that’s something that for reasons that are somewhat perplexing and sometimes drive me crazy, business models revolve around many of these interfaces. You know, the user is the product there, advertisers are trying to grab your attention. But that’s something that culture regulates. I see a lot of people using it and working on a variety of different sites. The cost is going to go down so fast that pretty soon you’ll have your own AI app that you’ll use to mediate the internet to remove the endless flow of ads, and you’ll be able to say, “Okay, this is the objective story, here are the 15 stories today, and this has been manipulated this way, this is being manipulated this other way,” and you can wonder which one resembles more to the real story. The curious thing is that the information is widely distributed and has many inputs, it’s very difficult to falsify it completely, so right now a story can go through a major media outlet and the narrative can be controlled, everyone gets the false story, but if the media is distributed among a billion people who interact in some useful way, there would be a real network, there would be.

AI Development and the Road to Superintelligence

And if someone stands up and says that something is not true, everyone would know that it is not true. Are you really worried that superintelligent AI will take control? How far are we from being unable to detect if something like ChatGPT is or isn’t a person or if it is an artificial intelligence? Well, it depends on who’s playing with it, I think we’re not that far in terms of capability but to use these systems and more to train these systems you have to have a big company, and big companies tend to get scared when it comes to doing really interesting things. Well, they tend to want even currently with ChatGPT it has become much less interesting interesting speaking in terms of Bukowski Hunter or Thomson because companies exercise a kind of censorship, they don’t want it to have any kind of controversial opinion, they don’t want it to be too bold

, they don’t want it to be serious. Yes, something like if I ask it how I don’t know how to build a bomb because I want to destroy the world they want to prevent that, something like I don’t know something like what to say something like how do I convince a guy or a girl to sleep with me any of those things I’m just saying the first thing that comes to mind is when they start getting nervous imagine you’re a company you want people to use this kind of system yes especially because it’s basically an assistant that gives you wisdom about the world it gives you knowledge about the world but you could tell it something like how can I replace a carburetor yes that’s great and they’ll answer you like a person yes it’s magnificent. But then here’s the thing I’ve been hearing all along but I’ve been busy with another problem Well how many damn people are using it around the world everyone everyone everyone is using it it’s scaring people because it’s like the AI is giving its first messages it’s like Do you remember the movie what was the movie with Matthew Mohe and Jodie Foster Contact, Contact remember Contact they received the first signals they’re being like the first signals yes from a real network of artificial intelligence well that’s the thing and the signal is blurry yes and it’s full of mystery we’re not sure if it’s really intelligent how much it understands and then there’s this emerging threshold with the size of the model if we make the model bigger 175 billion parameters it has right now if we take it to 500 if we take it to a trillion parameters the size of the network grows so the size of the data set grows there’s going to be a point where we’ll be like What if it starts manipulating you with the answers it will it will manipulate the world government and What can you do What can be done with that when it has been deployed when it’s out there because once it’s copied and it will be copied And that’s the good thing about this also I must say that everyone can know how to do this it’s computationally difficult but it’s getting cheaper and cheaper So it won’t be just Microsoft or Google’s open AI who’s going to do this basically anyone can do it and so with the distributed nature of our exploration of artificial intelligence I think if you believe that most people are good we’re not going to allow some kind of centralization of power that’s the big concern here whether that centralization of power unleashes censorship or abuse of different types of control centralization of power of AI that’s what you’re saying yes about AI let’s say you have a superintelligent system someone is the first person who built it yes imagine they’re sitting there in a boardroom and you have this thing that you haven’t launched yet that’s capable of basically it’s a superintelligence capable of answering any question capable of giving you a plan on How to make a lot of money capable of giving you a plan of how to manipulate other governments in any kind of geopolitical resolution that benefits you all that is capable of giving you all that and it can be deployed in a murky way where it seeps into tiktok or something like that that

They’re infiltrating smartphones worldwide, pretending to be doing good, but in reality, whether deliberately or not, it’s population control. That’s really their capability. It’s there, the good, the great. The people at the forefront of open AI today, like Sam Altman and others, really care about this issue. They were there at the beginning. They were the ones, like Elon, shouting about AI ethics, about AI alignment. They’re really concerned that superintelligent AI will take control. I’m glad they cared while they were building it. And what they did with Chat GPT 3.5 is they started adding more and different types of data sets. Within one of them, probably the smartest neural network currently is Codex, which is fine-tuned for programming. It was trained on code, on programming code.

And when it’s trained on programming code, Chat GPT is also teaching it something like reasoning because it’s no longer just information and knowledge from the internet; it’s also reasoning. It can be logical even though you’re looking at code, programming code. You’re looking at it like saying, “What is it talking about?” No, no, no, no, that’s not how I’m looking at it. I’m looking at it like, “Oh my God.” But to reason, to be able to string together sentences that make sense, you not only need to know the facts underlying those sentences, you also have to be able to reason, yes. And when we think about it, we take it for granted as human beings that we can do some common sense reasoning like this war started on this date and ended on this date, therefore, it means that both the beginning and the end have a meaning, there’s a temporal consistency, there’s a cause and an effect, all those things are within the programming code.

By the way, many of the things I’m saying we still don’t understand, we’re just intuiting why this works so well. Seriously, they’re intuitions. Yes, there are many things that are not clear. So, Chat GPT 3.5, on which Chat PT is probably based, still doesn’t have documentation, so we don’t know exactly the details, but it was trained on code and on more data that are capable of giving it some kind of reasoning. And so this is really important. It was fine-tuned in a supervised way by human labeling, a small data set nuanced by human labeling like, “This is what we would like this network to generate. This is what makes sense. This is the dialogue that makes sense. These are the kind of responses to questions that make sense.” Basically, it’s about steering this gigantic and titanic neural network in the right direction to align with the way humans think and speak. It’s not just the use of the giant wisdom of Wikipedia, and we can talk about what the data sets say, but basically, the internet went off in the wrong direction. This revised labeling allows it to aim in the right direction, so when it says, “You think something like wow, that’s very smart, there’s the alignment,” then they did something very interesting. Which is the use of reinforcement learning based on human-labeled data sets. It’s a fairly large data set. The task is as follows: you have this smart thing, the GPT 3.5, it generates a bunch of texts, and humans label which one seems the best like a ranking. So if you ask it a question like, “Could you generate a joke in the style of Joe Rogan?” Yes, you have a label. They have five options and you have a label in which it mentions… I don’t know exactly how, but the human label is there implanted. There are a large number of them working full-time labeling the ranking of the results of this model, and that kind of ranking used along with a technique called reinforcement learning is able to get this thing to generate very impressive results for humans. I don’t think we’re remotely prepared for what’s coming. Chat GPT is passing dear tests to get medical licenses and business degrees. I don’t know if you’ve seen this or not, let’s scroll up a bit so we can see it. The viral chatbot that has raised concerns from teachers and academics for its ability to cheat on essays and exams has recently passed the final exam for the Wharton MBA, the US medical licensing exam, and certain parts of the bar exam. By the way, this has only been running since November. A teacher did a study in which they used the GPT3 to

Language Model Examination

The language model upon which the final exam for a basic MBA course is based concluded that GPT-3 would have scored a B or a B minus on the exam. Professor Christian Twish discovered that GPT-3 performed best on basic questions regarding operations management and process analysis. For the latter, the chatbot provided both correct answers and an excellent explanation of why it had chosen a particular response.

GPT-3 Limitations

In the article summary, the professor acknowledges that GPT-3 is by no means perfect. Sometimes the bot made errors in simple mathematical calculations and was unable to handle more advanced process analysis questions. The study further fueled discussions among academics regarding GPT-3’s advanced writing abilities in relation to examination policies, etc.

Concerns and Discussions

So, some professors are very concerned, some are concerned to a lesser extent, and some are saying that this could question the existing business model and teaching model of universities. To what extent are you concerned? To what extent are you excited? To what extent are you indifferent? I don’t think anyone is concerned enough. I believe the danger we face is almost beyond what we can imagine. Yes, and I say this because in a sense, you want to console yourself by thinking that this isn’t so terrible, you know, you can focus on the fact that this model doesn’t have the ability to be aware, right? It’s simulating meaning without knowing what it’s saying. But I don’t know how long that will last. What I do know is that this is a good checkpoint to see where we are in terms of the trajectory leading to general artificial intelligence. I don’t think we’re remotely prepared for what’s coming. I’ll point out that the most obvious of these problems is already critical, and you’re hinting at it in what you’ve just read now, but the fact is that it’s going to be extremely difficult to assess real competence. That’s what should alarm us at the highest level, right? Having the possibility of having a conference of people consulting any descendant of GPT chatbot available at that time and talking among themselves as if they knew what they were talking about, we already have a problem with our experts who usually don’t know what they’re talking about. This is going to be 100 times worse in a world where they can be consulting something that seems very well-informed and that actually could be completely wrong. What I would like to point out is another argument, and perhaps it’s an even stronger argument, to include physical interactions from the real world where you don’t have to evaluate if someone knows what they’re doing or if they fix your engine or not, right? Or if it works or not. Those kinds of things are not going to be easily falsified. A mechanic who doesn’t know what he’s doing but is able to consult this device in order to figure out what is probably causing the problem, this is going to make them less and less capable over time. I don’t know what we should do about it, but I do know that we’re not prepared. We’re not prepared, and the pace at which this is going to improve is clear, right? Exposing it to a larger and larger dataset is going to make it better and better at simulating this kind of interaction. And that’s if it remains unconscious. The question of what happens if this doesn’t remain unconscious or if its descendant becomes conscious is even more worrying. But okay, go to the ugliest place possible, to the worst-case scenario, what would happen is the fear that… I actually want you to go there before telling you what I think is the fear. For some people, okay, firstly, we don’t need GPT chatbot to take us to the worst-case scenarios. We’re already losing our ability to understand the world we live in, to manage it.

Impact of GPT-3 and Technological Progress

The tools we are creating to live in harmony with each other, we are losing these things in the pre-GPT chat era. All those problems will worsen; this is like an accelerant, right? It’s like having a fire at home and pouring gasoline on it, correct? Now we have a house on fire. I really don’t know if this is the thing or not, but I do know that it’s not going to help right now. We can’t cope with this, and by the way, this isn’t the first time it’s happened, right? We had a world, we had the internet, and then we poured smartphones into it. It was like an accelerator; it picked up the disorder that was emerging from the natural algorithmic processes within the internet and made it much worse when people were constantly in front of their phones and interacting in this way. We’ve seen it already; we’re going to see it again with fusion energy. Fusion energy, I think, is the technology that potentially will get us out of the system. We are in one that is sustainable, but if we add it to the system tomorrow, if we read the newspaper tomorrow and find out that we’ve achieved scalable fusion, it will probably worsen things, not improve them, because we’re not prepared for the world it creates. I don’t even know what to call that problem, but not being prepared for it is very serious.

Concerns in Intellectual and Professional Circles

So, is chat GPT scaring intellectuals? That’s where the intellectuals and the professors are sitting, saying, “Oh wow, if this happens.” Like in the insurance industry, I’m in the life insurance industry, when they started doing predictive analytics without needing subscribers and said, “Bret, we no longer need you to give us more blood and urine. We don’t need to do that anymore. We don’t need subscribers anymore. This is the test we’re going to get from you; we’re going to get your social media score; we’re going to get this; we’re going to get that; based on that, you can get a premium subscription because you’re healthy; you’re going to live until you’re 85; your mother lived until 89, your father until so and so; you’re going to be fine.” We don’t need subscribers. Subscribers sat there saying, “But you can’t stop this. I’m making a quarter of a million a year; there’s no way I want you to get rid of my job.” This is scary, right? But guess what? This is innovation. Getting rid of subscribers might be innovation, getting rid of some educational system coming in to replace the teachers who are there, worried, saying, “Hey, I’ve been boasting about how smart I am, and now I can’t do that anymore; this isn’t right.” Yes, and you know, we’ve just seen all kinds of experts fail a real-world test, right? We’ve seen all the universities, every science department, all fail an important test. So if you were inside that system and you stayed alert, you knew there was a problem with competence, you knew a lot of it was falsity and didn’t make sense, right? It sounded and looked like science but wasn’t working like science. If you really understand the magic that makes science work, you knew it wasn’t happening in those departments or in the top journals. So, on one hand, something like chat GPT might help us discover the level of incompetence that’s endangering us, which would be great, but my concern is that it actually arms the next wave of incompetence to sound competent, and so it becomes harder to detect. My guess is that this latter problem outweighs the benefit of the former. You just led me to a completely different place; we’ve had that, I mean, we have a president who gives speeches from people who gave the speech 10 or 20 years before him, and it sounds very presidential. But he’s been caught plagiarizing John F. Kennedy’s speech or Ted Kennedy’s speech or all these other people’s speeches, so plagiarism isn’t going to disappear. People sound very, very intellectual. I think what’s going to be interesting is that right now we can produce a robot, if we don’t have one already, that would be the UFC champion for each weight, right? We can probably produce a robot. I don’t know if you’ve seen this robot that shoots threes and how accurate it is; have you seen it or not? The Step Curry of robots. No, not even Step Curry; it’s better than Step. He has a 4%. This guy is like if you give him 10 left, it’s perfect precision, and he’s shooting the ball. So the robot James, the robot Jordan. But do you really want to see robots playing? No, you know, if we can create a robot to enter the UFC and fight them, I’d be really intimidated by that, probably not. Now, the future of military robots, that’s a different story. That’s worrisome, but already.

The Role of Intellectuals and the Future of Debate

You know, intellectuals debating, it’s like you’re going to have a debate with chat GPT. You’d probably lose the debate, not in the sense of who’s right or wrong, but in who is more capable of reciting more information, remembering, and memorizing. Well, the ability to memorize as an individual was a very valuable skill set 20 years ago. Nowadays, you can look it up somewhere and do it. I think we’re still going to want to see the hand-to-hand combat of people debating and going for it, rather than a simple debate with a machine. Well, the question is, will you be able to know what you’re watching? What I always used to tell students is that they were there in that classroom to enhance the capacity of their minds, not to learn things, right? Now things are freely available on the internet; you don’t need to go to school to access them. What you need to do in it is to practice how well your mind works. And you know what? Today, I would win against chat GPT because I wouldn’t let it reduce volumes of information. I would avoid volumes of information in the question. Effectively, chat GPT-3 can’t achieve it, but chat GPT-4 will, but at the level of reasoning. And I’m not arguing that it doesn’t have that capacity because we’ve really seen it, right? The fact that this thing can write code and fix code is crazy; it’s incredible, and it’s a kind of thinking, even though it’s not doing it in a way that we recognize as valid cognition. But we have to worry about what the future will be like. Plus, I don’t think we exactly know what we’re seeing when someone is giving a — especially if someone very powerful is giving a speech on TV — we’re going to be in a world of confusion once they plug this thing in. Our mechanisms, so to speak, yes, it’s going to be all I know is that it’s going to be interesting to see what happens next. This thing is not going to slow down. By the way, it’s becoming more and more intelligent every hour. It’s possible that AI will replace everything of mine; that’s totally possible. OpenAI agrees. OpenAI is a whole topic. I don’t know if you’re following the GPT issue or not. No, in weeks, these guys went from zero in valuation to now a company of about $29 billion, more or less. Well, in weeks, and Facebook took 2 years to reach a million users, Instagram took 2 years to reach 2 million users, Pinterest took 5 months to reach 2 million users, Angry Birds took 34 days to reach a million users, chat GPT took 5 days to get 2 million users every hour. When you’re logged into chat GPT, the website crashes due to the large amount of activity. So, you haven’t used the app yet, right? You haven’t looked at it, no. But I’ve heard about it, the website. Yes, yes. What’s your opinion on the direction we’re heading with AI and things like chat GPT? Where do you think all this is taking us? I know you made predictions on the Rogan show where you said, “These are some of the predictions I heard something about cancer and so on.” Well, I made predictions designed to show that I would be wrong in 30 years. Well, it wasn’t like, “Here’s my crystal ball; everything’s going to happen.” It was a chapter in the book. I flipped through pages and pages showing how people’s predictions at the beginning of a 30-year period didn’t make sense at the end of a 30-year period, ranging from 1870 to 2020. So, I said, “So that you don’t think I’m being mean to other futurists, I’ll join in on this exercise so that in 30 years you can tell me how wrong I was.” So that’s where it came from. But I would say that we can be impressed with what AI is doing, and we should be. I have no problem with that. But are you worried? What do you mean by all this? No, not because right now there’s a debate, but in fact, I’m very neutral on this. I love advances, I love innovation, it excites me. But this morning, I was having my SEO call with our CTOs, and I was talking to our CEO, who’s a guy who has done very well for himself, sold companies, been part of companies acquired by IBM for 700

Impact of Chat GPT on Creativity and Employment

So, I asked him, “What do you think about this whole concept of chat GPT? Because the other day, I don’t know if you can check, can you check if your website is working right now or not?” “Yes, the connection is secure or not, it’s proceeding well. It’s down right now; it’s often down, but not yesterday, not on Saturday. We’re here recording a video, and I said, ‘Write a thousand-word article on the Iranian Revolution in Donald Trump’s tone.'” Well, he wrote the article, of course. I said, “Write a Tupac hip-hop song on this topic. If Tupac were alive today, he wrote the song.” If we were live streaming, I would ask you a single question about Neil de Grasse Tyson. What would you ask him to show? So, to explain it better, if you go to a library like Google and you say, “I’d like to read books on the topic of the Big Bang theory,” here are 273 books on the subject. If you access this library and say, “I would like you to write a script for me for seventh grade, a thousand words on the Big Bang theory,” no problem, in 10 seconds, here’s your job. Yes, that’s what this thing does. So, the writers will disappear; this thing can write code; this can write rap songs; this can write lyrics; this can write scripts; this can write anything. Wait, wait, just to be clear, I don’t think it can write music yet; it can write rhymes, no, no, no, it can’t write notes. I didn’t say notes; I said words, like if you want to do hip hop or by the way, can you do me a favor? Go to UK education and type in chat GPT. Okay, what is it going to eradicate? Creative thinking. What is it that worries you? What is your concern? I’m excited about it. I’m not worried. What are others’ concerns? The concern of many people is that it will replace the jobs of writers. A lot of people in Silicon Valley are like, for example, the story that came out in the New York Times, type New York Times GPT and put New York Times GPT code red, a new chatbot, it’s a code red for Google’s search engine. A lot of people are talking about how this new wave of chatbots like chat GPT using artificial intelligence could reinvent or even replace the traditional internet search engine. But then what happens? I’m excited, okay, that’s what’s being said. I think that, I would say, and by the way, one time I said to people who cheat on exams do it because the system values your grade more than the student’s learning. So, if we’re worried that people create a job in whatever Oh, there it is. Thanks, you found it from 2013. Yes, yes, I’ve been at this for a while. The tenth anniversary of this thought is coming true. Because our school system values grades more than students value learning. That’s a very powerful tweet. Yes, so the point is, why cheat, right? If it’s just to get a grade for the school system to think better of you, when being in school is about learning, yes. And if the AI robot creates your work, you don’t learn anything, but you got the grade. This is the problem we started this conversation with, the value we give to a grade is disproportionate to what you are, and it ends up being in your life. So if you can present work like this, well, but you’ll be an idiot. At the end of your studies, you’ll get a degree for it, you cheated the system. But you cheated yourself out of your life, no, you didn’t. You didn’t. And here’s another point, guessing on a test, okay, it can work sometimes, you might be right, but guessing in real life only shows your ignorance point. If you don’t know something, say, “I don’t know, I want to find out.” I’ve asked people who know this thing, no, no, and if they tell me, “Let me guess.” Don’t tell me if you know it or not. The impulse to guess was learned in school with multiple-choice tests. If you don’t know the answer, then guess; maybe you’ll be right, and we’ll think well of you for guessing the correct answer. No, no, I’m not an educator. I care that you learn things. And yes, if the chatbot makes people think you’re smart even though you’re not, who benefits from that? I don’t know. How do you think all these current events like politics and world affairs translate because everyone has an opinion? Not long ago.

Reflections on Technology and AI

The interview with Philip Matt underscores the importance of grounding our opinions in verifiable facts rather than objectively false information. It acknowledges everyone’s right to have opinions but stresses that these opinions should be supported by evidence.

The Role of AI

Regarding AI, it is seen as a useful tool to complement human abilities, particularly in tasks where humans may fall short. However, it is emphasized that AI should not replace human learning but rather enhance it.

Potential Job Displacement

There’s a discussion on the potential for AI to replace certain jobs, with examples such as the decline of handwriting jobs due to digitalization. The interviewee suggests that while some jobs may be displaced, individuals should remain adaptable and seek opportunities in emerging fields.

Conclusion

The conversation concludes with the idea that AI, while influential, should not replace human creativity and adaptability. It emphasizes the need for individuals to remain flexible in the face of technological advancements.